Prediction of Hyper Thyroid Disorders using Classifier algorithms in Data Mining

Abstract

Thyroid disorders in women are known as one of the most common diseases. The thyroid gland regulates the metabolism of the body and its development. It also secretes several hormones, such as Calcitonin, Thyroxine (T4), Tri-iodothyronine (T3). Women of any age can be affected by thyroid issues. Women are more likely to have thyroid disease than men. Such symptoms include hypothyroidism, hyperthyroidism, thyroiditis, goitre, thyroid nodules, thyroid cancer. There are also risks if the thyroid condition is untreated and unrecognized. It could be recognized using data mining algorithms,. The proposed work is to build a model that can diagnose the probability of hyperthyroidism with reasonable precision in patients. Naive Bayes, Random Forest, J48, PART classifier algorithms are used to detect the hyper thyroid problem. Simulation studies were performed for experimental data sets sourced from UCI machine learning repository using these classifiers. The performance of these classifiers is analyzed on various performance metrics, such as Precision, Accuracy, F-measure, and Recall. Accuracy measured over true and false classified instances. PART outperforms with the highest accuracy of 97.99% comparatively other classifiers.